Athlon 64 Venice core power picture explored


— 12:58 AM on May 3, 2005

Michael Schuette over at LostCircuits has written a fascinating new article on the Athlon 64 "Venice" core. He attempted to isolate the power going to the CPU on his test system in order to measure CPU-only power consumption. He seems to have done so, but at a cost:

It is possible to measure the current output directly at the Vcore output, however, the high switching frequency of the VRM makes the hookup somewhat hazardous for the VRM itself since phaseshifts induced by extra capacitance can easily blow up the entire circuitry – as we found out the hard way. Not a pretty sight.
Doh!

When he got new hardware, his work apparently continued, and the resulting article sheds new light in several areas. For instance, his look at the connection between CPU temperatures and power consumption is interesting. I've observed this effect in testing for my own reviews. (Effectively, our power consumption results come from the lower end of the graph; we don't allow time for much time for thing to heat up before taking our reading.) He's also quantified performance per watt (scroll down) in several common benchmarks. In Doom 3, the Venice achieves more than double the performance per watt of the older Clawhammer core. Very impressive.

Oh yeah, and he got his 3800+ to run stable at 2.8GHz, too, which also isn't bad.

I have a Venice core in house here for testing. We should have some results from it before too long, but some even more interesting CPUs may be first in line.

 
   
Register
Tip: You can use the A/Z keys to walk threads.
View options

This discussion is now closed.